If you follow the CoreOS documentation and try to get a Vagrant cluster up and running - the etcd cluster is not created by default.
After cloning the Vagrant CoreOS repo, cd into the cluster folder and run vagrant up. This will start 3 core-os clusters numbered core-01 to core-03. You can ssh into them by running :
vagrant ssh core-01
In my case the IP addresses of those coreos clusters are:
- core-01: 192.168.65.2
- core-02: 192.168.65.3
- core-03: 192.168.65.4
etcd -s 192.168.65.2:7001 -cl 0.0.0.0 -c 192.168.65.2:4001 -d nodes/node1 -n node1&
This will start the etcd server (port 7001) and a client (port 4001), store the files (configuration, logs) in nodes/node1 and name it node1. The -cl 0.0.0.0 option is to enable etcd to listen to both internal (0.0.0.0) and external IPs.
etcd -s 192.168.65.2:7001 -c 192.168.65.2:4001 -C 192.168.65.3:7001 -d nodes/node2 -n node2&
This is pretty much similar to the parameters passed to etcd on core-01 except the -C option which points to the IP address of core-01
etcd -s 192.168.65.4:7001 -cl 0.0.0.0 -c 192.168.65.4:4001 -C 192.168.65.3:7001 -d nodes/node3 -n node3&
Note: Daemonizing the etcd daemons the way I did is definitely not the way to go. I've seen examples where they are executed within a docker container but this will do for a quick test run.
Now you have an etcd cluster. To test that they all see each other you can run:
curl -L http://192.168.65.3:4001/v1/machines
The output should list the three clusters:
http://192.168.65.2:4001 http://192.168.65.3:4001 http://192.168.65.4:4001
You can access the etcd dashboard by pointing your browser to: (on your host machine)
You can now follow the rest of the documentation and publish messages to etcd, read them on all the clusters.